ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments
Not on the map. Why?
Abstract (arXiv)
Scientific code repositories encode decades of human knowledge in executable models, methods, and tools. Yet fragmented toolchains, implicit domain conventions, and specialized correctness criteria make this knowledge difficult to convert into reliable learning experience-a challenge we call the scientific experience bottleneck. We introduce ScienceIDE, infrastructure for turning the world's scientific code into programmable environments for scientific agents. Guided by expert-defined scientific cases and acceptance criteria, agents transform repositories into executable environments that support task generation, execution, and scientific verification. These environments provide a shared foundation for supervised fine-tuning, reinforcement learning, and evaluation. Using verified interaction trajectories, we train PhAI-IDE-72B, PhAI-IDE-9B, and PhAI-IDE-4B. The model family shows gains in held-out scientific-code repair and across selected general-purpose benchmarks in code, reasoning, and knowledge, providing evidence of positive transfer from scientific experience to broader capabilities. ScienceIDE lays the foundation for an integrated workspace for agent learning and scientific practice, making humanity's scientific software a shared substrate for developing scientific intelligence. Code: https://github.com/aitofound/ScienceIDE
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 85 of its 139 references that are in the map. First posted on arXiv 16 Sep 2026.
Builds on
85 of 139 references.
References · 85
- 2022Training language models to follow instructions with human feedbackcited by 15,712
- 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
- 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
- 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
- 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
- 2017Deep reinforcement learning from human preferencescited by 3,846
- 2023Let's Verify Step by Stepcited by 3,598
- 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,193
- 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,806
- 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,748
- 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,678
- 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,081
38 more
- 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 1,980
- 2023WebArena: A Realistic Web Environment for Building Autonomous Agentscited by 1,908
- 2024SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineeringcited by 1,628
- 2022Solving Quantitative Reasoning Problems with Language Modelscited by 1,359
- 2024τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domainscited by 1,280
- 2025Understanding R1-Zero-Like Training: A Critical Perspectivecited by 1,276
- 2023AgentBench: Evaluating LLMs as Agentscited by 1,211
- 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,193
- 2020ALFWorld: Aligning Text and Embodied Environments for Interactive Learningcited by 1,143
- 2023GAIA: a benchmark for General AI Assistantscited by 1,084
- 2024The AI Scientist: Towards Fully Automated Open-Ended Scientific Discoverycited by 1,002
- 2024OpenHands: An Open Platform for AI Software Developers as Generalist Agentscited by 956
- 2022STaR: Bootstrapping Reasoning With Reasoningcited by 895
- 2025AlphaEvolve: A coding agent for scientific and algorithmic discoverycited by 763
- 2022FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operatorscited by 748
- 2022Galactica: A Large Language Model for Sciencecited by 736
- 2024Executable Code Actions Elicit Better LLM Agentscited by 537
- 2019AI Feynman: a Physics-Inspired Method for Symbolic Regressioncited by 536
- 2019Leveraging Procedural Generation to Benchmark Reinforcement Learningcited by 478
- 2024MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineeringcited by 429
- 2025Agent Laboratory: Using LLM Agents as Research Assistantscited by 405
- 2021Flow Network based Generative Models for Non-Iterative Diverse Candidate Generationcited by 384
- 2025The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Searchcited by 375
- 2017Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equationscited by 362
- 2025Accelerating scientific discovery with Co-Scientistcited by 353
- 2023Beyond Human Data: Scaling Self-Training for Problem-Solving with Language Modelscited by 325
- 2018TextWorld: A Learning Environment for Text-based Gamescited by 304
- 2024Training Software Engineering Agents and Verifiers with SWE-Gymcited by 300
- 2025PaperBench: Evaluating AI's Ability to Replicate AI Researchcited by 286
- 2022GraphCast: Learning skillful medium-range global weather forecastingcited by 261
- 2025SWE-smith: Scaling Data for Software Engineering Agentscited by 255
- 2025SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolutioncited by 246
- 2024FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning in AIcited by 227
- 2024DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Datacited by 225
- 2020PettingZoo: Gym for Multi-Agent Reinforcement Learningcited by 209
- 2024ScienceAgentBench: Toward Rigorous Assessment of Language Agents for Data-Driven Scientific Discoverycited by 208
- 2020The NetHack Learning Environmentcited by 196
- 2021Benchmarking the Spectrum of Agent Capabilitiescited by 193
50 most cited of 85 All 85 (JSON)
Built on by
Citing papers per year: 2026: 2.
2 papers.